Hacker News Reader

Comment by blueprint | Hacker News Reader

Such a good comment. But I have a question for you. You said this is 'one of the major interpretations' of QM. But if the hypothetically greater future data is not observable by anything outside of the universal wavefunction, then is it not the case that the entire system is evolving without external observation, i.e., the system itself does not have to be able to verify that it "knows" what outcome will occur for it to be able to generate new data by a causal mechanism. So it sounds like randomness is not proven by what you said is definition, alone. Is this what you meant by it being "one of the" interpretations of QM?

View on HN

Well I define random to mean whenever P(X|Y) is neither always 1 nor 0, or to put that in causal modal terms:

given an exhaustive description of all possible causes of Y, say X, then if X causes Y, *NECESSARILY* Y given X

causation is a form of necessity, necessarily Y occurs or it doesnt

In this case P(Future|Past) and P(Past|Future) are somewhere between 0 and 1. This isnt consistent with causation.

ie., All possible information about all possible causes of Y (the future), say X (the past) are insufficient to gaurentee any given Y (any particular future).

The "interpretation" element of QM is to say that we are really in this case, ie., that we have X = all possible information.

There are several ways of denying that P(Y|X) in the case of QM should be read this way. You can say X is incomplete, either by saying there is non-local information we don't have, or global information we don't have. Of the "global" kind, which of many worlds we are in is a kind of global ignorance.

So that if we had, P(Y|X, which-world) then we'd observe P(Y|X, whichworld) = either 1 or 0 --- as with classical mechanics

P(CatDead|PoisonBox, worldA) = 1

P(CatDead|PoisonBox, worldB) = 0

etc.

and, eg., P(F|G,M,m,r) = 1 if F=GMm/r^2

what do you mean by the term information? My question is, rather, whether a system needs to be able to know it knows that information for it not to still be causal. Because it doesnt observe its whole wavefunction

I dont understand your use of "knows", knowing isnt part of the problem.

Imagine, for simplicty sake, a world is a series of four switches which we can describe by 0 (off),1 (on).

Eg., world A at time 0 is 0000, at time 1, 0001, ...

Now a change from 0000 to 0001 is causal if in every physically possible world 0001 follows 0000, ie., P(0001|0000) = 1

Now suppose that we enumerate every possible world, ie., all the worlds containing only this kind of switch, and only four of them.

Now if 40% of the worlds had 0000->0001 and 60% had 0000->1000, then P(0001|0000) = 0.4

The question is how can it be that the fourth switch behaves randomly? Ie., without any necessary flip from 0 to 1. Nothing at all can cause that switch to flip, by definition, the world contains nothing else in it.

What property does the switch have such that it turns on 40% of the time? NONE.

Consider two resolutions to this problem: bite the bullet and say the switch is "miraculous in some sense" in that it "just turns on" sometimes; OR, say that this thought experiment is physically impossible and we're missing information.

Eg., that there's a hidden switch: 10000->10001 always, and 00000->11000 always, we just can't see the first switch, so it looks random.

"Imagine, for simplicty sake, a world is a series of four switches which we can describe by 0 (off),1 (on)."

If that's what the world is then there's no "we"

Indeed, by definition, that world consists of four objects with four distinct properties, ie., being on or off.

Then where is the time bit?

> if the hypothetically greater future data is not observable by anything outside of the universal wavefunction

If there is a universal wavefunction evolving deterministically what would it mean to talk about the future holding more data than the past? There would be a direct one-to-one correspondence between the future states and the past states.

No, I don't think that's correct. I'm not a professional but.. despite unitarity and maybe CPT symmetry, I think that a given "past" (by which I think we mean the universal wavefunction at some t?) has no way of recording what its own state is. If measurements were to be done of subsets or the whole of the wavefunction, then it would, I imagine, contribute to the state of the wavefunction so as to evolve it. But keep in mind as well when we say "data" we're talking about state which can be measured, aren't we? Maybe some state can't be observed; and, the "universe" also definitely limits its own certainty anyway.

> when we say "data" we're talking about state which can be measured, aren't we?

I don't know! If the "data" is about the state of the universe (i.e. the wavefunction) the "data" doesn't really change because of unitarity.

You may be talking about something else - but it's not clear what.

what time it is here changes from one moment to the next. and there are also horizons. you dont have all the data about things yet you operate. just trying to help

but we already knew yesterday what time would be tomorrow - there is no new data in that sense

obviously not true - in so many ways. what exactly in a wavefunction knows whether it will encounter other wavefunctions? i think what you're claiming leads to many contradictions.

The only thing that I claim is that hartator wrote about the future holding more data than the present and past and it’s unclear what that would mean in a deterministic universe represented by a pure quantum state under unitary evolution.

What is data? How do you establish -given a wave function- the amount of data that it holds? What are you claiming exactly?

But our universe, as well as a theoretically deterministic universe, necessarily, certainly contain many mixed states.

And, I didn't use the term data first. But when I used that word I was trying to transmit the meaning of "content", "substance", or character, nature, or what exists in the matter/issue. What is in the principle of a system's operation. How do I establish given a wave function the amount of data that it holds? A wave function doesn't exactly contain data. You either prescribe a wavefunction, or try to model one based on hypothetical constraints, or you can attempt to guess what an existing actual wavefunction was through measurement. As I mentioned I think when we say data we're usually talking about what is measured. But there is something more that exists than what is measured. Something that you could say is invisible doesn't not exist if it determines everything else that exist. If you understand everything about an object or a system you can prescribe an exact wavefunction. But we do not understand everything about what is in the objects for which we have tried to prescribe wavefunctions. What am I claiming exactly? I was not claiming anything, although now I have made a claim at the start of this comment; but I was asking a question as to whether 'mjburgess' was merely phrasing his original comment to make it seem to the careless reader or undiscerning eye that 'If the future "holds more data" than the past then it's random with respect to the past, by definition. This is one of the major interpretations of QM.' was actually "If the future "holds more data" than the past then it's random with respect to the past, by all possible definitions (implicitly). This a conclusion finally reached by physicists after having studied QM, rather than merely one possible interpretation given the best theory available so far.' but they indeed did not write that.

Reply on news.ycombinator.com